Sakana AI Hires Jürgen Schmidhuber as Chief Scientific Advisor

Sakana AI has hired Jürgen Schmidhuber — the researcher the Tokyo-based company calls the "father of modern AI" — as its Chief Scientific Advisor. In that role, he will help lead the company's newly formed RSI Lab, which focuses on recursive self-improvement: AI systems that keep developing themselves.

The announcement, reported by The Decoder, connects one of the field's most cited foundational researchers to one of its most aggressive research agendas. Sakana AI says Schmidhuber's ideas from the 1990s have already shaped its projects, including the Darwin Gödel Machine.

Who is Jürgen Schmidhuber?

Schmidhuber is a German computer scientist whose work in the 1990s laid groundwork for several strands of modern machine learning. The Decoder's reporting on the hire identifies him as an inventor of deep learning and world models, and notes his connection to technologies that underpin today's chatbots. Sakana itself describes him with a superlative label: the "father of modern AI."

That framing is more than marketing. Schmidhuber's decades-old research on learning systems and self-referential improvement directly anticipates the problem Sakana's new lab is now trying to solve: how to build AI that improves itself, rather than waiting on human engineers for every advance.

What is the RSI Lab?

The RSI Lab — the name stands for recursive self-improvement — will operate under Schmidhuber's scientific guidance at Sakana AI.

Its mandate, as described in the source report:

• Research AI systems that develop themselves continuously
• Build on the self-improvement concepts Schmidhuber explored in the 1990s
• Extend work already underway at Sakana, such as the Darwin Gödel Machine

The Darwin Gödel Machine is the concrete evidence that this agenda predates the hire. According to The Decoder, Schmidhuber's ideas from the 1990s already shaped that project, which takes its conceptual lineage from self-referential systems he studied decades before large language models became commercial products.

Why the hire matters

Recursive self-improvement sits at the contested center of AI research. If AI systems could reliably improve themselves, the pace of capability gains could decouple from the pace of human engineering effort — a prospect that drives both commercial ambition and safety debate across the industry.

By hiring Schmidhuber as Chief Scientific Advisor, Sakana AI is pairing that ambition with a researcher whose ideas the company says its existing projects already build on. The move signals that Sakana intends to compete on frontier research directions — not only on products — and that it sees self-improving systems as its differentiator among AI labs.

The hire also reflects a broader pattern in the AI market: laboratories are racing to attach foundational researchers to their most speculative programs, on the theory that credibility with the research community attracts both talent and capital.

What comes next

Sakana AI has not announced a timeline for the RSI Lab's first results, and The Decoder's report does not specify additional staffing for the lab beyond Schmidhuber's advisory role. What is clear is the direction: Schmidhuber will help steer research toward systems that improve themselves, with the Darwin Gödel Machine as the existing proof of concept. Whether recursive self-improvement can move from a 1990s research idea to a working engineering practice at Sakana will be one of the company's defining tests.